[HALF-solved] Cant Play Complete Shutdown Crash- AMD

[Nice]

Edit/update on fix: Ive since managed to play the game for hours but still had 1 crash ever since, the 'solution' here is hardly a solution but an workaround via gimping and limiting the gpu itself. Whatever is causing this shouldnt as most other games dont bypass the gpu termal throtling. - btw i recall playing badly optimized demos that did the same but their full release didnt... i suspect some source codes and engines allow such fine tuning in talking to the gpu directly they can ask for too much or cause a loop that ramps up the temp unecessarily.

Oh my, i was expecting issues with connecion and server, not this... My game isnt just crashing to desktop- its forcing a complete system shutdown. I cant be no more then 1
minute on the world before it happens.

Cpu: Ryzen 3600x
GPU 5700xt
Windows 10
32gb ram
4k monitor

Everything is up to date, i even DDU my gpu drivers and reinstalled just to be sure.
Tried everything, even everything set to performance... since theres no fullscreen and TSR was relative to my 4k monitor ive tried changing the desktop resolution, to 1440p then 1080p...

I suspected of TSR causing some bug, so i tried 1080p native, also crashed. Then 1080p performance (absolutely horrible btw)

Every single attempt my machine was hard shutting down.
I suspected of overheat, and maybe that was it, but i find it extremely unlikely - even at 4k TSR balanced i was at 60 fps with my gpu saying it was at 60-70% usage. It only reduced in usage.
I RUN GAMES IN MORE DEMANDING ENVIROMENTS all the way 90+- gpu, at high settings, for hours... something isnt adding up.

Tried some time later with the machine completely cooled off AND air conditioning on my room turned on - running on everything performance/lowest... and it didnt took any longer for the pc to crash into shutdown.

WHAT IS GOING ON?

Please help, tell me theres some ini setting or something i can try. Maybe settings on AMD adrenaline?(ive run with pretty much everything turned off)

Makes no sense, ive been gaming the entire week from light to heavy gaming, including unreal 5 with lumen, absolutely nothing causes that.

I wish i hadnt spent so long in character creation like twice, now my refund window is even lower... But dammit i dont want to refund, not in principle- i want it to work ♥♥♥♥♥♥♥♥. Damn i was so excited, what a terrible first contact...

PS: also why the hell theres no dedicated fullscreen option? Ive had scenarios in the past where games would peform much better or worse depending on screen setting, some performing better at dedicated fullscreen, borderless or windowed. Besides having one less thing to test i think its terrible usability sopping the player from changing the base resolution like it happens with borderless fullscreen.

[Rogue]

Did you overclock, CPU or GPU or RAM? Tone them down. E.g Curve optimizer, -30 lower to -20.

[Nice]

I never touched anything like that, ive always been scared of doing something wrong- outside of tweaking gaming profiles its everything out of the box. Except fans wich i tweaked once awhile ago (but just found out adrenalin ignored/reseted those settings probably because of the DDU).
Unless you count ram "TPM" (the equivalent on amd asus motherboards) but i never had any issue with that

Speaking of wich back when i built my pc ive read some people were underclocking or undervolting either the 3600 or the 5700, for reasons i cant fully remenber but in 5700 case the gpu launch was rocky, with non optimized drivers. But i never went that route and some driver updates later never had game crashs or such so i never bothered...

Would underclocking or undervolting for Nightingale in particular help? Is there a simple noob way of doing that?

[saice]

I had this issue in the stress test.

The only fix that worked for me was going into the files and flagging the exe for compatibility mode Win8

[Rogue]

Try default gpu settings in adrenaline, no enhance sync, etc. Basically all the toggle is to the left. Also cap FPS to 60 in-game and use Xess.

[Nice]

Originally posted by saice:

I had this issue in the stress test.

The only fix that worked for me was going into the files and flagging the exe for compatibility mode Win8

Just tried it didnt work for me, shutdown again.

Originally posted by Rogue:

Try default gpu settings in adrenaline, no enhance sync, etc. Basically all the toggle is to the left. Also cap FPS to 60 in-game and use Xess.

I was already running everything off/left after the first shutdown. I will give Xess another try but i doubt it will help, on my initial fiddling with settings ive tried it and even on performance was much worse FPS wise (sub 60, i think it was 45fps or so vs 60 from TSR, on the starting cavern where i didnt crash). If the issue is overheating alone my guess Xess will crash faster.

But i will try, just not sure if today- im getting tired of what might have been like the 10th or 12th crash-shutdown-restart in short intervals.

But i saw some hope. Before the last crash i also tweaked the least scary part of the adrenalin tuning and made the fans 100% quickly... and this time it took like 6-7 minutes before shutting down. The suggested 'windows 8 compatibility mode' was also on, and i was also trying some launch option fix i found - '-psomaxthreads=8' on steam launch options... im not sure if any of those impacted the results

So im more inclined to believe its overheating- but i dont understand it; Ive ran heavy demanding games that when i checked were that hot or even more, to the point hours later i could feel how hot the air around my case was without touching it. Its like something here overrides or ignores whatever safety measures on the gpu/driver to prevent that (like throtling)

I havent quite managed to see at wich point in temperature the crash happens, ive been 'alt-tabing' the amds adrenalin overlay to check temps, but id say its crahsing around 87-88 or 90...

I will try some more tests tomorrow and probably try undervolting (gotta learn how to do that first)

[Arc]

Spontaneous reboots are always a hardware or a system problem.
In rare cases can be caused by unstable power grid, though if it happens regularly after several hours, it's likely overheating, OCP or OVP kicking in.

Also, there is no Fullscreen Exclusive in modern DirecX 12, it's not needed, and the DX12 games that have it, emulate it.

[Nice]

I may have solved it
Last night i managed to run for a bit over an hour- i dont know if its definitive or if it was going to crash later at some point.
What worked for me:
- was undervolting my GPU- i went with AMDs own 'auto undervolt'.
- running as windowed instead of borderless window may have helped. It was one of the tests i did where it took longer for shutting down.
- it was late at night and the overall ambient temperature was lower (and i live in a hot country)

99% sure its the gpu hotspot going overboard and failing to kick in throtling etc- but what doesnt make sense is that i can run heavy games on mid day summer without air conditioning without any shutdowns. I know the hotters ones and because in such scenarios (hot days, hot games) i can noticeably see the throtling ramping up, with a very linear progression of fps lowering over time.

Brings to memory the New Worlds beta that burned the top tier nvidia cards- it was some fault on their hardware but it was only kicked by some bizarre code mess from New World devs

Originally posted by Arc:

Spontaneous reboots are always a hardware or a system problem.
In rare cases can be caused by unstable power grid, though if it happens regularly after several hours, it's likely overheating, OCP or OVP kicking in.

Also, there is no Fullscreen Exclusive in modern DirecX 12, it's not needed, and the DX12 games that have it, emulate it.

Not reboots. The system doesnt restart by itself, full shutdown, very likely the gpu safety, wich made me think of overheat.
But thats the thing- it doesnt happen regularly. I can be gaming for hours on end on high demanding games and nothing happens.

Ive also got suspicious of Unreal 5, maybe something to do with its implementation, but i cant be sure. Ive also run some other Ue5 games (through demos only) but wich didnt cause that, even for hours. Heck i recall one demo in Ue5 being so terribly optimized (no fault of unreal, it was a messy alpha) it ran like 15 fps at most on my machine, i was afraid it would blown up/shutdown but... nothing. Maybe it would, but it lasted longer the Nightingale did before the auto-undervolt

I dont know if its some mistake on AMDs end but heck its only happening with Nightingale. I personally didnt want to undervolt my card... i mean ive considered it, some people talk of no performance loss with lower temps or improved performance and lower temps, but i worry id mess up and AMDs auto-undervolt is know to be bad- some lower temp with lower performance...

[0===]]\'_\'_\'_\'_\'_\'_\'_/]

hello have you discovered where the problem comes from because I have the same problem impossible to find where it comes from it's a first For me however I have played games much more complex in terms of graphics 😱

[VYM]

Try ThrottleStop, click FIVR and limit all turbo groups cores @32 or lower. It will limit your CPU power and avoid overheating.
I limited my i7 14700 @42 or 4200mhz to make temperature not over 65-70C.

[Nice]

Originally posted by 0===\'_\'_\'_\'_\'_\'_\'_/:

hello have you discovered where the problem comes from because I have the same problem impossible to find where it comes from it's a first For me however I have played games much more complex in terms of graphics 😱

Yes, it was what i described last post. I half solved it tough- i can now play for hours BUT i had 1 crash since, and it was on a cold morning... Im convinced that overheating is PART of the problem, but whats making the amd overheat seens to be some particular bug and i heavily suspect of the effects of the game. I notice some throtle (for like 3 seconds before the crash) then the crash when i was QUICKLY breaking rocks one after another

Anyway in short, anything you can do to force the gpu to underwork, limit itself or apply heavier countermeasures/safety. In my case i used amds auto undervolt:
- Open Radeons Adrenalin
- Performance tab
- Tuning
- Choose undervolt (accept the warning if you didnt)

But other capping solutions can also solve- if you ever overclocked your gpu disabling it at least for Nighingale, what VYM just said about ThrottleStop probably works too...

Ive also found these 2 guides on making the game itself less demanding wich definetly help:
https://steamcommunity.com/sharedfiles/filedetails/?id=3164854981
https://steamcommunity.com/sharedfiles/filedetails/?id=3166075259

I didnt need to, just undervolting was enough (but i will disable fullscreen optimizations just to be sure)

PS: before i undervolted running on windowed mode made the game take longer to crash BUT reduced performance/fps, ever since i undervolted didnt need to. Now that i think about it windowed may have helped precisely because its worse- since it made the game underperform it took longer to stress the card.

[Kai]

Originally posted by Vincer:

Edit/update on fix: Ive since managed to play the game for hours but still had 1 crash ever since, the 'solution' here is hardly a solution but an workaround via gimping and limiting the gpu itself. Whatever is causing this shouldnt as most other games dont bypass the gpu termal throtling. - btw i recall playing badly optimized demos that did the same but their full release didnt... i suspect some source codes and engines allow such fine tuning in talking to the gpu directly they can ask for too much or cause a loop that ramps up the temp unecessarily.

Oh my, i was expecting issues with connecion and server, not this... My game isnt just crashing to desktop- its forcing a complete system shutdown. I cant be no more then 1
minute on the world before it happens.

Cpu: Ryzen 3600x
GPU 5700xt
Windows 10
32gb ram
4k monitor

Everything is up to date, i even DDU my gpu drivers and reinstalled just to be sure.
Tried everything, even everything set to performance... since theres no fullscreen and TSR was relative to my 4k monitor ive tried changing the desktop resolution, to 1440p then 1080p...

I suspected of TSR causing some bug, so i tried 1080p native, also crashed. Then 1080p performance (absolutely horrible btw)

Every single attempt my machine was hard shutting down.
I suspected of overheat, and maybe that was it, but i find it extremely unlikely - even at 4k TSR balanced i was at 60 fps with my gpu saying it was at 60-70% usage. It only reduced in usage.
I RUN GAMES IN MORE DEMANDING ENVIROMENTS all the way 90+- gpu, at high settings, for hours... something isnt adding up.

Tried some time later with the machine completely cooled off AND air conditioning on my room turned on - running on everything performance/lowest... and it didnt took any longer for the pc to crash into shutdown.

WHAT IS GOING ON?

Please help, tell me theres some ini setting or something i can try. Maybe settings on AMD adrenaline?(ive run with pretty much everything turned off)

Makes no sense, ive been gaming the entire week from light to heavy gaming, including unreal 5 with lumen, absolutely nothing causes that.

I wish i hadnt spent so long in character creation like twice, now my refund window is even lower... But dammit i dont want to refund, not in principle- i want it to work ♥♥♥♥♥♥♥♥. Damn i was so excited, what a terrible first contact...

PS: also why the hell theres no dedicated fullscreen option? Ive had scenarios in the past where games would peform much better or worse depending on screen setting, some performing better at dedicated fullscreen, borderless or windowed. Besides having one less thing to test i think its terrible usability sopping the player from changing the base resolution like it happens with borderless fullscreen.

New AMD driver updates, probably fixes it.

Also some aspects of the game drops your framerate even if the game can handle it, like chopping trees when your camera is up close to it. It seems to 'explode' something which causes frames to stall.

I tweaked the settings to force tessellation off, just in case the game is secretly 'exploding' millions of tiny triangles every time a tree is chopped.

[0===]]\'_\'_\'_\'_\'_\'_\'_/]

Vincer ,thank you for your response, I think I will wait for an update from them because apparently we are not the only crash victims 🤣

[Nice]

Originally posted by Kai:

New AMD driver updates, probably fixes it.

Also some aspects of the game drops your framerate even if the game can handle it, like chopping trees when your camera is up close to it. It seems to 'explode' something which causes frames to stall.

I tweaked the settings to force tessellation off, just in case the game is secretly 'exploding' millions of tiny triangles every time a tree is chopped.

Good to know! I had changed adrenalin settings for the game for tesselation to be handled by the application (just in case amd optimized was helping crash), just turned it to override->none. I will later update if i noticed any changes

[Mixologic]

Same issue here. AMD 5700xt, 4 hard shutdowns in the character creator, 1 once I finally got past it. Turned on undervolt and it seems to be working... so far.

[? ? ?]

btw i am not sure a 5700xt is a4k card XD.... why u paly wiht a 5700xt on a 4k monitor whats the reason.

If u go down to qhd i looks ♥♥♥♥♥♥